WO2021187704A1

WO2021187704A1 - System for creating parking map and confirming vehicle location in parking lot, using deep learning and rtt signal

Info

Publication number: WO2021187704A1
Application number: PCT/KR2020/013825
Authority: WO
Inventors: 김경식
Original assignee: 김경식
Priority date: 2020-03-15
Filing date: 2020-10-09
Publication date: 2021-09-23
Also published as: KR20210116780A

Abstract

The present invention relates to a system for guiding parking and confirming vehicle location, using deep learning, wherein a system server (11) and a plurality of user terminals (12) are connected by using a network, a user terminal (12) transmits, to the system server (11), data sensed at a set interval by using an altitude sensor (25), an acceleration sensor (26), and a gyro sensor (27), the system server (11) then stores the data in a movement path DB (42), and a virtual map creation module (43) comprises a pre-learned deep learning network and creates a map via a method of predicting, on the basis of the data stored in the movement path DB (42), a path on which a vehicle travels, a space expected to be a parking space, and a space that is neither a path or a parking space, and expressing same on a map, and then stores the map in a virtual map DB (47), and receives data from a plurality of users and learns repeatedly.

Description

Parking map production and vehicle location verification system using deep learning and RTT signal

The present invention relates to a system for making a parking map using deep learning and RTT signals and for checking vehicle location in a parking lot.

The present invention relates to a system for making a parking map using deep learning and RTT signals and for determining the location of a vehicle in a parking lot. Many drivers have had the experience of being lost because they could not find the location where they parked in large parking lots such as large shopping malls or airports. The parking lot side attaches a sign with color and number on the pole to make it easier to find the parking lot, but if you park for a long time, you may forget the parked sign. In some cases, you can find the location through sound by pressing the car key, or check the parking location at the parking payment machine or parking confirmation machine installed on each floor. In many cases, it was difficult to find a parking location with the help of a device such as a cell phone because GPS does not work in the parking lot.

In this regard, Republic of Korea Patent Registration No. 10-2059849 'Intelligent Parking Management Method and System' takes a picture of a vehicle that wants to enter a parking lot, recognizes the vehicle number and vehicle type information, and maps the parking path to a parking space based on the IoT sensor. This user terminal is provided for the purpose of only providing guidance to the available parking space, and does not include a configuration for finding a car when getting back into the car after parking.

Even if you try to find the location of the car when you return to get in the car after doing business in the building after parking, it is difficult to find the location of the car because it is not possible to determine exactly where the current location is or where it is parked. It is necessary to provide a separate parking map for each parking lot and provide a separate means for determining the user's location, but since it costs a lot for each parking lot, it cannot be forced on the parking lot owner, and the parking lot owner does not feel the need to use such a device. can't

In particular, since GPS cannot be used in underground parking lots, it is difficult to make a map in the parking lot using GPS. The portal company makes a map and provides it, but for the map of the parking lot, the portal company sometimes needs to take a car elevator to get to the parking lot. Since there is no need to create such a map in the parking lot, the building owner often has to bear it, but the building owner also makes a map of the parking lot and does not feel the need to install and maintain an additional device in the parking lot to use it. Also, since there is no other way to use the map in the parking lot, even if it is made, only one page of guidance is created and provided, and it is rarely provided separately online.

In particular, because there is no location information by GPS, Wi-Fi, BLE, etc. that can confirm the location in the underground parking lot, existing positioning technologies such as GNSS, WPS, RFID, and beacon cannot confirm the location of the vehicle in the parking lot, and It takes a lot of time and money to receive or produce maps of the underground parking lot in each building, and it is very difficult to commercialize it because if you try to install a separate Wifi AP or beacon in the parking lot, you have to go through a separate consultation process for each building. .

The present invention has been devised to overcome the above disadvantages, and uses a sensor basically attached to a user terminal, but accumulates information received from a large number of users, and determines the aisle and parking location for each parking lot from this information. An object of the present invention is to provide a system for guiding into a space or facilitating finding a parked location.

In order to achieve the above object, a system server 11 and a plurality of user terminals 12 are connected using a network to provide a parking guidance and vehicle positioning system using deep learning, and the user terminal 12 transmits the data sensed every predetermined time using the altitude sensor 25, the acceleration sensor 26, and the gyro sensor 27 to the system server 11, and the system server 11 stores it in the movement path DB 42 And the virtual map making module 43 predicts a ball that is not a parking space, and a passageway through which a car travels, a space expected as a parking space, and a passage based on the data stored in the movement route DB 42 including the deep learning network learned in advance. Thus, a map is produced by a method of displaying it on the map and stored in the virtual map DB 47, so that data from multiple users can be received and repeated.

The user terminal 12 further includes a camera 21 so that the system server 11 receives the image from the user terminal 12 together and stores it in the image information DB 45, and the OCR reading module 22 provides text in the image. can be recognized and stored in the character recognition DB (46) and stored together with the corresponding location in the virtual map DB (47).

The facility recognition module 44 may recognize the facility in the image information DB 45 and store it in the virtual map DB 47 together with the corresponding location.

The user terminal 12 may receive information about the speed and direction of the vehicle from the vehicle OBD module 13 and store it in the movement path DB 42 of the system server 11 .

The GPS sensitivity check module 24 may check the sensitivity of the GPS and provide it to the system server 11 to check whether the vehicle is located outdoors, in a building, or in a tunnel.

The available parking space prediction module 48 checks the vehicle parked in the position predicted as the parking space in the image information DB 45 from the license plate of the vehicle, and learns the vacant time for each parking space according to the time to determine the parking prediction DB ( 49), and whenever there is a request from the user terminal 12, information on the available parking space can be provided to the user terminal 12.

The parking guide module 51 may provide a path to the available parking space.

The vehicle location finding module 50 can use the information obtained from the user terminal 12 to determine the location of the vehicle using the GPS 23 outdoors and verify where it is located on the virtual map of the parking lot in the building. .

The parking location identification module 52 may determine a movement route in the parking lot after the user gets out of the car, and guide how to move from the departed route to the parked location when the user leaves the pre-written virtual map. .

The parking location identification module 52 may predict where the user is located on the virtual map and provide it to the user terminal 12 when the user provides any one of a picture and an image of a location where the user is located in the parking lot.

Find ai,bi,ci that satisfies the following equation, but from 3 or more equations,

By obtaining the normalization correction coefficient ui

The data sensed by the sensor can be corrected by the following equation.

(S is the sensing value Si,t of a specific sensor, t is the time, data from the first sensed time to the end of the sensing (p), S'i,t is the normalized sensing value using Si,t, sigma i-1 is the standard deviation of the i-1 data, sigma i is the standard deviation of the i-th data set Si,av is the sensing average value of the I-th data, Si-1,av is the sensing average value of the i-1 data)

With a1=a2=a3=..an, b1=b2=b3=...bn, c1=c2=c3=...cn, for the same sensor, n correction factors are all the same, so one value can be used. have.

Without the need to configure a separate system for each parking lot, it accumulates information received from the user terminal and learns parking spaces and passages to provide drivers with a virtual map that is almost identical to the actual map. It has the effect of making it easy to know where you parked and where you parked.

1 to 4 are diagrams showing an embodiment according to the present invention;

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings. 1 shows a block diagram of a system according to the present invention; The system server 11 according to the present invention is connected to a terminal installed in a vehicle or a terminal carried by the user through a network, and the terminal installed in the user or vehicle is collectively referred to as the user terminal 12 .

The user terminal 12 includes a camera 21 and an OCR reading module 22 for OCR (Optical Character Recognition), a GPS 23 and a GPS sensitivity check module 24, an altitude sensor 25, and an acceleration sensor 26. . You can have all the sensors that a general mobile phone has, such as a gyro sensor 27, a brightness sensor 28, a hall sensor 29, a temperature/humidity sensor 30, a geomagnetic sensor 31, and an RGB sensor 32. There is no need for a sensor, and if necessary, a fingerprint recognition sensor, a proximity sensor, etc. may also be used. In addition, the sensor may further include a sensing value check module 33 for checking a change in the value sensed by the sensor.

Among them, the altitude sensor 25 senses the pressure of the air and can detect a slope or a downhill, and can be used to detect the number of floors.

The acceleration sensor 26 measures the moving speed and acceleration in a three-dimensional coordinate system, and measures the moving speed/acceleration of the user terminal.

The gyro sensor 27 may be used to measure a moving direction in a three-dimensional coordinate system to measure a moving direction of the user terminal 12 .

The brightness sensor 28 detects the brightness of ambient light, and is used to automatically adjust the brightness of the display by sensing the ambient brightness in a general mobile phone.

The hall sensor 29 is a sensor used to sense the strength of the surrounding magnetic field, and is often used to check whether the cover is closed or not.

The temperature/humidity sensor 30 refers to a sensor that measures the temperature and humidity around the user terminal 12 .

The geomagnetic sensor 31 refers to a sensor that detects the direction of the Earth's magnetic field. Detect the N and S poles of the compass.

The RGB sensor 32 refers to a sensor that detects the color density of ambient light, and a smartphone with an RGB sensor is used to correct the color of the display according to the surroundings and the density.

In addition, proximity sensors and fingerprint sensors can be used as needed, but mainly, altitude sensors, acceleration sensors, gyro sensors, brightness sensors, geomagnetic sensors, etc. can be used to detect the direction and speed of movement, and other sensors can be used auxiliary. can

In addition, the user terminal 12 may receive and use information about the vehicle's moving speed and the vehicle's direction from OBD (On-board Diagnostics) attached to the vehicle. The vehicle OBD module 13 and the user terminal 12 may be connected via Bluetooth or the like. OBD is attached to vehicle management by vehicle manufacturers. Recently, OBD information is directly received from vehicle and used for vehicle management.

In addition, when a terminal supporting RTT and a plurality of WIFI APs supporting RTT are installed in a parking lot, the location can be more accurately determined. Using RTT is to send an FTM (fine timing measurement) signal for distance measurement to the WIFI AP at a specific location, and when the user terminal sends a response signal to the FTM signal sent by the AP, the AP sends and receives the FTM signal from the time the user terminal and the AP are received. The distance between the two is measured. At this time, after the user terminal receives the signal, the response signal is sent back to the AP. Except for the time processed by the terminal, the time taken from the AP to the user terminal and the time taken from the user terminal to the AP are used. to calculate the distance. When calculating the distance, wireless signals travel at the speed of light, so if you can calculate the time it takes to send and receive signals, you can also calculate the distance by assuming that these signals have traveled the same path. Figure 2 shows these contents. In FIG. 2, using the actual signal movement time (t4-t1)-(t3-t2) and the speed of light, the distance between the user terminal and the AP can be calculated.

In this way, the distance from a specific AP to the user terminal can be measured, and when multiple APs are used, a more accurate location can be calculated. That is, when the distance from the specific AP to the user terminal is obtained, the user terminal may be located on a circumference forming a predetermined distance centered on the specific AP. However, if the distance from another AP to the user terminal is obtained in the same way, two circles can be drawn with the two APs as the center and the distance to the user terminal as the radius. location of the terminal.

In order to use this method, the location of the AP must be specified, but even if it is indoors, if the AP signal is measured at one location and the same AP signal is measured again after moving a certain distance, the AP will measure the distance measured at each point as a radius When you draw a circle with , it is located at the point where the two circles meet. If the AP signal is measured several times after moving a certain distance again, the location of the AP can be quickly obtained even if the location of the AP is unknown. Fig. 4 shows these contents. In FIG. 4, three APs are used, and two APs are also possible, and the accuracy increases when three or more APs are used.

In order to determine the user's location, two or more APs transmit wireless signals at the same time and obtain the difference in the delay time of the propagation time arriving at the terminal, thereby obtaining the difference in the distance between the user terminal and each AP. Since the number of attempts for location detection using a user terminal is not fixed, it is easy to find the location indoors by repeatedly making these attempts when the user moves.

In addition, since it is repeatedly performed by a plurality of users, the location of the AP can be determined by repeated calculation. When the location of the AP is confirmed, the location of the user terminal can be easily obtained when there are two or more APs.

The user terminal 12 receives the location information of the GPS 23 where the GPS 23 is operable, stores the movement history in the user terminal 12 and transmits it to the system server 11 . So, in an outdoor parking lot with GPS, it is possible to determine the parking space and movement route only with GPS without the need of other sensors.

In an area where the GPS 23 does not work, the GPS sensitivity check module 24 checks the sensitivity of the GPS to check whether it has entered a building or underground. Although it is not a parking lot, such as a tunnel or a building, a place where the sensitivity of GPS is low can be confirmed by the GPS sensitivity check module 24 checking the sensitivity.

When entering a building or underground, the sensitivity of the GPS decreases, and values of other sensors other than the GPS 23 are importantly used where it is recognized that the vehicle has entered the parking lot. In particular, the movement path of the vehicle in the parking lot is drawn by the acceleration sensor 26 , the gyro sensor 27 , the altitude sensor 25 , the brightness sensor 28 , and the like. Fig. 2 shows an example of this. The left drawing of FIG. 2 shows a form in which the movement path of the vehicle is grasped using a sensor at regular time intervals in the vehicle after entering the parking lot. Although not all parking locations have been identified, the more these moving routes are stored, the more clearly the passage, the parking space, and the other spaces can be distinguished. The drawing on the right of Fig. 2 shows the movement route on the map of the actual parking lot. If you have a terminal that supports RTT and an AP that supports RTT is installed in the parking lot, the location of the user terminal is determined using RTT. In the parking lot, the vehicle can become a wall, but if the AP is installed on the ceiling, there is no delay due to the vehicle.

In addition, the camera 21 and the OCR reading module 22 recognizes and reads the text in the parking lot and transmits it to the system server 11 along with the movement route. The user moves around the parking lot and parks when a vacant seat appears. The place where the user parks is recognized as a sure parking space. is recognized as a space with a high probability of being a parking space, and the space next to the passage is more likely to be a parking space.

The moving path is transmitted to the system server 11 and stored in the moving path DB 42 in the parking lot DB 41 . It is not necessary for the user terminal 12 to wander around the parking lot in order to draw a movement route, and the user terminal 12 simply enters the parking lot as usual and parks in an empty space. Data that can distinguish between and parking spaces are accumulated.

The movement route transmitted from the user terminal 12 is stored in the movement route DB 42, and the movement route is provided from a plurality of user terminals 12 using the corresponding parking lot instead of one user terminal 12, and the movement route DB ( 42) and using the accumulated movement path, the virtual map making module 43 draws a virtual indoor map and predicts the actual parking area by using the parking information. The image information synced with time along with the sensed information transmitted from the user terminal 12 is stored in the image information DB 45, and the facility recognition module 44 receives the image information from the image information DB 45 from the image of the parking lot pillar, the circuit breaker, Recognize parking facilities such as crosswalks, stairs, and doors. The recognized text is stored in the character recognition DB 46 together with the recognized position.

In particular, the position where the user terminal 12 moves is predicted as a passage, and the finally parked space is determined as a parking space, and other spaces are predicted as a space with a high probability of being a parking space. In addition, by recognizing the image taken by the camera 21 and the license plate number in the image, if the vehicle is correct, the space can be predicted as a space with a high probability of being a parking space. The space where the facility is located is also recognized as a space where parking is not possible. The facility recognition module 44 distinguishes a parked vehicle from a separate facility, considers the moving time of the vehicle and the location of the camera, checks where the corresponding space is located on the map, and determines whether the space captured by the camera is a parking space let it do This determination is made by a pre-trained deep learning network.

In the case of an rtt-supporting terminal, the location of the terminal can be identified while exchanging a signal with an rtt-supporting AP at regular intervals. In fact, when determining the path, wall, parking location, etc., the exact location cannot be grasped with just one or two measurements, so the results of measurement by several people are taught. . In the case of using the wifi-rtt signal, it is possible to obtain a more accurate location.

The virtual mapping module 43 uses information sensed from a camera, a sensor attached to the user terminal 12, and information transmitted from the vehicle's OBD to create a path and a parking area of a parking lot, an area that is neither aisle nor a parking area, etc. It learns that the area where the user actually parks the vehicle is learned as an area with a high probability of being a prescribed parking area, and the area where other vehicles are parked several times through the camera is also learned as an area with a high probability of being a parking area.

The virtual map produced in this way is stored in the virtual map DB 47, and every time new data comes in, it learns again to reflect changes such as areas that are closed to prevent parking by the parking manager or areas that have been closed and reopened. to learn

In the past, maps were mainly made using GPS, so it was difficult to create a map in a space where the GPS did not work, and even if a map was made, there were not many ways to use the map. In particular, if there is a change in the space in the parking lot, the map may be useless if not updated. The cost for map management can be greatly reduced.

In addition, after the map is completed, the available parking space prediction module 48 uses the image information DB 45 and the movement route DB 42 and, if the virtual map is completed, the virtual map DB 47 according to time. By learning the empty parking space, the parking space and the empty space are predicted according to time and stored in the parking prediction DB (49).

When the virtual map is created, it is stored in the virtual map DB 47, and when the user enters the parking lot, the vehicle location identification module 50 determines which parking lot has entered from the GPS information and stores the virtual map of the parking lot in the system server. Imported from the virtual map DB 47 connected to (11), the available parking space prediction module 48 secures the information of the parking space predicted to be available for parking, and the parking guide module 51 provides the path to the space to the user terminal. (12) can be sent. The vehicle location finding module 50 uses the GPS 23 to determine the location of the vehicle in the outdoors and verifies it using the information obtained from the user terminal 12 where it is located on the virtual map of the parking lot in the building.

The parking location identification module 52 detects a movement route in the parking lot after the user gets out of the car, and guides how to move from the departed route to the parked location when the user leaves the pre-written virtual map. In other words, it indicates where the parked location is at the parking lot exit. At this time, it is possible to guide the pillar number near the parking position as well, and it can also guide the number of floors. In addition, the location where the user parks is stored in the parking history DB 53 so that it can be easily found even after a time has elapsed.

If there is one exit in the parking lot, this is sufficient, but if there are several entrances to the parking lot, the user location identification module 50 detects the user's location, and the camera Through recognition, the camera recognizes the passage in the parking lot, and the location where the user is most likely to be in the current parking lot is found and transmitted. When the sensing values of the acceleration sensor 26 , the gyro sensor 27 , and the altitude sensor 25 are added, it becomes easier to find the user's location in the parking lot. Or, if the WIFI -rtt signal is available, a more accurate location determination is possible.

Since the signal sensed from the user terminal 12 is sensed and transmitted from the user terminals owned by a large number of people, there may be a slight difference in the sensed value depending on the age or performance difference of each terminal. In this case, only the values of some devices may be excessively reflected. To prevent this, the sensed values are corrected and used so that all sensed values are reflected at a similar rate.

In the same sensor, in Si, av, i means the data of the entire path transmitted in the i-th, and av means the overall average of the corresponding sensing value to the data in the i-th. In other words, normalization is performed on the basis of the data transmitted immediately before, and the number for normalization is obtained with several data sets that came in immediately before.

In Si,t, t refers to data from the first sensed time to the end of sensing (p) in time, and the sensed value normalized using Si,t becomes S'i,t. Si-1.av represents the overall average of the i-1th transmitted sensed values, and sigma represents the standard deviation. Sigma i-1 is the standard deviation of the i-1 data, and sigma i is the standard deviation of the i-th data set.

If the sensed value is acceleration data, if person A enters the parking lot, parks, and then transmits the sensed data, it becomes the (i-1)th, and person B enters the parking lot, parks, and transmits the sensed data. becomes the second Equation 1 is used to find ai, bi, ci that satisfy the expression in the i-th , and the following correction coefficient ui is obtained using the approximate value found in Equation 1. How many formulas are used in the front to find the variables ai, bi, and ci to find the correction factor, and how many formulas are used depends on the calculation ability that can be done at once.

In Equation 1, it is possible to find a, b, and c if there are 3 or more data to find the variables a, b, and c. a, b, and c can be found as values that satisfy the entire data, and for 3 or more i, a separate value can be found for each i. That is, one a like a1=a2=a3=... It is possible to obtain and apply to the whole, then obtain it again at a certain period, or to obtain all a for each and use it.

Equation 3 is an expression that is normalized using the normalization variable Ui. The i-th data and the i-1 data are logically unrelated data. . However, as the iterative operation is made, even if there is little logical correlation, each data is normalized as it affects each other.

This method reduces the calculation time of the system server 11, and although there may be a difference in the values that are initially normalized, as data accumulates, calculations are performed multiple times, resulting in similar results to those calculated at once. In particular, for learning, the normalization step has to be repeated over several generations, and deep learning learning converges as time goes on because these repetitions are repeated several times.

(Explanation of symbols)

11: system server 12: user terminal

13: vehicle OBD module 21: camera

22: OCR reading module 23: GPS

24: GPS sensitivity check module 25: altitude sensor

26: acceleration sensor 27: gyro sensor

28: brightness sensor 29: hall sensor

30: temperature / humidity sensor 31: geomagnetic sensor

32: RBG sensor 33: sensing value check module

41: parking lot DB 42: moving route DB

43: virtual map production module 44: facility recognition module

45: image information DB 46: character recognition DB

47: virtual map DB 48: available parking space prediction module

49: parking prediction DB 50: vehicle location identification module

51: parking guidance module 52: parking location identification module

53: Parking history DB

Claims

As a system for making a parking map using deep learning and checking the location of a vehicle in a parking lot,

The system server 11 and a plurality of user terminals 12 are connected using a network, and the user terminal 12 uses an altitude sensor 25, an acceleration sensor 26, and a gyro sensor 27 every predetermined time. The sensed data is transmitted to the system server 11, and the system server 11 stores it in the movement path DB 42.

The virtual map making module 43 predicts a ball that is not a parking space, but a space that is expected to be a parking space and a passageway through which a car travels based on the data stored in the movement route DB 42, including the deep learning network learned in advance. A system for creating a map by displaying it on the map and storing it in the virtual map DB (47), using deep learning to repeatedly learn by receiving data from a large number of users
According to claim 1, wherein the user terminal (12) further comprises a camera (21), the system server (11) receives the image from the user terminal (12) together and stores it in the image information DB (45), the OCR reading module ( 22) recognizes the text in the image, stores it in the character recognition DB 46, and stores it in the virtual map DB 47 along with the location.
According to claim 2, wherein the facility recognition module 44 recognizes the facility in the image information DB (45) and stores it together with the location in the virtual map DB (47), parking map production using deep learning and vehicle location in the parking lot verification system
The method of claim 3, wherein the user terminal 12 receives the vehicle speed and direction information from the vehicle OBD module 13 and stores it in the movement path DB 42 of the system server 11, using deep learning. Parking map production and vehicle location confirmation system in parking lot
The parking map using deep learning according to claim 3, wherein the GPS sensitivity check module 24 checks the GPS sensitivity and provides it to the system server 11 to check whether the vehicle is located outdoors, in a building, or in a tunnel. Manufacturing and vehicle location check system in parking lot
The method according to claim 5, wherein the available parking space prediction module 48 identifies a vehicle parked in a position predicted as a parking space in the image information DB 45 from the license plate of the vehicle, and determines the vacant time for each parking space according to time. It learns and stores in the parking prediction DB 49, and whenever there is a request from the user terminal 12, information on the available parking space is provided to the user terminal 12. verification system
The system according to claim 6, wherein the parking guide module 51 provides a path to the available parking space, making a parking map using deep learning and checking the vehicle location in the parking lot.
The method according to claim 7, wherein the vehicle positioning module 50 uses the information obtained from the user terminal 12 to determine the location of the vehicle using the GPS 23 outdoors and where it is located on the virtual map of the parking lot in the building. Parking map production and vehicle location verification system using deep learning to verify
The system according to claim 8, wherein the vehicle location identification module 50 uses two or more WIFI-rtt support APs to determine the location, using deep learning to make a parking map and check the location of the vehicle in the parking lot
The method according to claim 9, wherein the parking location identification module 52 identifies a movement route in the parking lot after the user gets out of the car, and when the user leaves the pre-written virtual map, how should the user move from the departed route to the parked location? Parking map production using deep learning and vehicle location confirmation system using deep learning
The method according to claim 10, wherein the parking location identification module 52 predicts where the user is located on the virtual map and provides it to the user terminal 12 when the user provides any one of a picture and an image of his or her location in the parking lot. Ha, parking map production and vehicle location verification system using deep learning
12. The method of claim 11, wherein ai, bi, ci that satisfy the following equations are obtained, but from three or more equations,

By obtaining the normalization correction coefficient ui

A parking map production and vehicle location verification system using deep learning that corrects the data sensed by the sensor according to the following equation

(S is the sensing value of a specific sensor Si,t to t is the time from the first sensed time to the end of the sensing (p), S'i,t is the normalized sensing value using Si,t, sigma i-1 is the standard deviation of the i-1 data, sigma i is the standard deviation of the i-th data set Si,av is the sensing average value of the I-th data, Si-1,av is the sensing average value of the i-1 data)
13. The method of claim 12, wherein a1=a2=a3=..an, b1=b2=b3=...bn, c1=c2=c3=...cn, for the same sensor, all n correction coefficients are Parking map production and vehicle location verification system using deep learning that uses one same value